Evidence-based treatments (EBTs) for depression are often not available to persons needing them; this is particularly true of psychotherapies. Even when available, EBTs are often poorly delivered at less-than-optimal quality. High direct and indirect costs also limit the availability of EBTs. Together these barriers contribute to suboptimal treatment of depression in the community. In a preliminary step toward addressing these quality shortcomings, we propose to conduct a blended efficacy- effectiveness randomized controlled trial (RCT) of high fidelity, Internet-delivered cognitive behavioral therapy (CBT) for depression, extending our previous research to maximize treatment availability and quality as well as to reduce costs. Over a 36-month recruitment period, we will enroll 1,800 adults seeking care for depression from 3 rural healthcare clinics, 3 safety net federally qualified healthcare centers (FQHCs), and 2 non-profit HMOs. Participants will be randomized to: (a) a treatment as usual (TAU) control condition, typically antidepressants and/or psychosocial services; (b) TAU plus Pure self-help Internet CBT for depression, consisting of access to the Internet site without any contact with therapists; (c) TAU plus Guided self-help Internet CBT, consisting of access to the Internet site plus brief, periodic telephone contacts with therapists; or (d) a Stepped-Care Internet CBT condition, starting with TAU + Pure self-help CBT and progressing to Guided self-help CBT if adequate progress is not observed early on. Participants will be followed for one year. The primary hypothesis for which the study is powered is that Guided self-help CBT will result in greater depression symptom improvement than Pure self-help CBT. We also expect secondary analyses to reveal this pattern of results: Guided CBT > Pure CBT > TAU. We will conduct cost-effectiveness analyses (CEA), as we project substantial differences in the direct costs of each study arm. We will also examine TAU healthcare utilization (medications, visits, etc) from electronic medical records (EMR), billing systems, and participant report. We hypothesize that cost per depression free days (DFDs) and quality- adjusted life years (QALYs) will be lowest for Pure CBT, relative to Guided and Stepped-Care CBT and TAU. We also hypothesize that cost per unit of improvement in QALYs and DFDs will be better for Stepped-Care compared to Guided CBT. Additional aims include exploratory examination of secondary outcomes, and predictors and moderators of outcomes among the interventions. We also will collect quantitative and qualitative data on patient, provider, and organizational factors that may impede or facilitate implementation of these interventions, to help prepare for future dissemination efforts. Finally, in this reapplication we have added a non-research Reach Estimation Phase to better estimate acceptance and retention rates under conditions that closely match real-world dissemination.